Currently, the only measure of performance is your time and a grading of puzzles across four levels. A new rating index is now calculated for all puzzles, roughly ranging between 900 to 1600. This attempts to provide a much finer-grained indicator of puzzle difficulty although, as you well appreciate, this is not an exact science.
The actual score is calculated when you complete the puzzle according to the following formula:
Score = Rating + TimeBonus - Accuracy - SolveAids
where:
- Rating = the rating index assigned to each puzzle. This index attempts to measure the difficulty of the puzzle by examining at each stage of the solving process how many cells can be solved by logic relative to the number of unsolved cells. It also looks at what techniques are required to solve those cells. Those that rely on more sophisticated techniques greatly increase the rating index. The current difficulty level (Easy, Medium, Hard, Tough) is only based on the techniques required to solve and you may now find that some Easy puzzles have a higher rating than other Medium ones or similarly for Medium/Hard/Tough.
- TimeBonus = extra points for completing the puzzle quickly. In fact, time is still the most important factor determining your score. The time bonus is computed by taking the average time per difficulty level divided by your time and then multiplied by the rating index. For example, the average time for Medium puzzles is currently set at 12 minutes (720 seconds) so if you complete a puzzle with a rating of 1100 in 6 minutes, you will score a time bonus of (720/360 * 1100), ie 2200 which is twice the rating for that puzzle. If you completed the puzzle in 20 minutes (1200 seconds), your time bonus would be (720/1200 * 1100) ie 660.
- Accuracy = the number of errors you make. For the first cell that you enter an incorrect value, you lose 30 points, for each subsequent error the points lost increases by 10. So you would lose 40 points for the second error and so on. Note that pencil marks do not affect the scoring in any way.
- SolveAids = each time you use a hint or clue or check your input. You lose 40 points for each hint, 50 for each clue and 20 each time you check your input.
The parameters of the formula will most likely be refined based on empirical data collected as well as user feedback. Please feel free to comment here.
great job, i love the new scoring system. keep up the good work.
ReplyDeleteOut os curiosity, are you in the future going to accumulate the scores and rank the scores in some kind of index?
ReplyDeletemkilian
Yes, the idea is to create a personal ranking based on average points scored so you can measure your own performance and also relative to others. It will also be used in the challenges.
ReplyDeletej'aime beaucoup jouer au sudoku sur ce site. merci! thanx
ReplyDeleteThe scoring system is great! Will be more exciting to know how others are scoring on a given puzzle - say if you could display the top and the 10th (or 101st) best score as well.
ReplyDeleteThe daily challenges do show how you score relative to others for the same puzzle. Currently this is measured in time but will be changed to points soon.
ReplyDeleteI like the new scoring system but would like to make a couple of suggestions. First, if you select "Auto Show Errors" in your preferences, you get the equivalent of an input check, but no scoring penalty is shown. Second, it would be interesting if you were penalized for guessing, i.e. Entering a correct value in a position that has two or more possibles.
ReplyDelete"Auto Show Errors" should definitely count as an input check each time (I thought it was). Thanks for pointing that out - now fixed.
ReplyDeleteMy initial plan for the scoring was to penalize for guessing correctly when not logically solvable. I actually had this logic implemented in the Sudoku Challenge initially - if 5 or more occurrences of this was recorded, I would display message "irregularity detected" and not post the entry. I got a few nasty responses to this saying they were not cheating. I am sure in these cases they probably were not willingly cheating and thought each step was a logical entry (easy to miss a possible number and think there is only one). So to avoid any dissension, I have not implemented this.
There's something going on with scoring. I just did three in a row, two "hard" and one "tough". All three showed 0 errors but over 50 checks each. The score, as a result was very low. What's happening?
ReplyDeleteI think there's some kind of a problem. I just solved the "hard" puzzle in 11 minutes with no errors. The score was 1800 or so and it said that there were 53 checks. What does that mean?
ReplyDeletePlease read my previous comment above. The reason for all the checks is that you have "Auto Check Input" option set so it checks after every input. The "check" penalty should have been happening from the beginning but was an oversight on my side. I will make this issue clearer in the documentation.
ReplyDeleteis 3000 a good rank? is 3500 better or worse
ReplyDeletewhat is a good score for a game? is 4606 in the top 100 for a game?
The higher the rank the better (as it is based on score). I will be publishing top ranking players and also the average rank soon, after I have gathered more stats. As for top scores for a puzzle, I will be changing the daily challenge to scores so you will be able to see what you score relative to all others who did the same puzzle. I expect some very high scores based on the incredible times some players seem to be able to post.
ReplyDeleteIs there a penalty for useing the pause? Some people will print the page and pause Then fill in the game after the printed paper game is solved. That seems to be unfair. Also the fastest typist will always have the best score/ranking not the fastest/best player. I would prefer pause or print to get a penalty. If you play a game more then once in a given time frame, say an hour or two, should have a penalty since memory will be in play not the deduction to solve the game.
ReplyDeleteDerek, I love the updates! Excellent work. One thing I would like to request is that the game page would show if we're logged in or not. Since there's no indication, I log in everytime, but I'm not sure how persistent the cookie is, so I don't know if I should do it once a day or one a month. Thanks!
ReplyDeleteI don't think users should be penalized for using the pause - it does not give an advantage unless you specifically set out to cheat (like printing puzzle and solving). If that's what a user does, well they are the only losers (there are no prizes or money for a better ranking). In specific challenges, I agree that the pause should be disabled (as it is for the daily challenge). That said, I think printing a puzzle should make it ineligible for a score.
ReplyDeleteAs regards logged in or not, it does show your name next to the "Play" button if you are logged in. If not, it will show a login link. But now that there are a lot of registered users, I will be adding a login prompt at the top right of each page. It will show your name if logged in.
This afternoon (1/25/10) I played the game 114660002 (H). I got to a point where I didn't know the answer. Selecting "Clue" responded with "Cell can be solved with OnePlace (advanced)".
ReplyDeleteI just reviewed all of the tutorials and I didn't see a "OnePlace" tutorial. Will "OnePlace" be added as a tutorial?
The message is misleading and needs to be changed. "OnePlace" is a basic strategy (see first tutorial) and it is being displayed because it was the last strategy used to solve that particular cell (after one or more advanced strategies were used). The message should really spell out all the strategies (but there is a lack of space).
ReplyDeleteDoes the rank index exclude the score from the daily challenge?
ReplyDeleteDaily challenge scores were not updating the stats (and therefore rank index) but that has now been changed.
ReplyDeleteI've played several different levels of games on several different days. Each time I complete a game and the "Well Done!" box comes up, the High Score is listed as 5874. Each time it is 5874, What does that mean?
ReplyDeleteIs 5874 the score of the person who got the highest score on that game? or is it some other value. If its the highscore for the game, I would expect each game to have a different score.
The High Score is the highest score that you personally have ever scored on a puzzle. It is not the high score for that puzzle (as new ones are generated all the time and you may be the only one to ever do that specific puzzle). The only puzzles that have all the scores recorded are the daily challenge puzzles and you should do these if you are interested in how you score relative to others for the same puzzle.
ReplyDeleteI would like to see a ranking available when using the possibles easy feature. For me, it's more fun to get to the meat of the puzzle so one can try the more advanced techniques earlier.
ReplyDeleteIn the regular games I enjoy seeing the rating of each game. Is there any way of displaying the rating of the daily challenge games?
ReplyDeleteSteven
I will add the puzzle rating to the message at the end of the daily challenge.
ReplyDeleteSometimes I forget to change pen for the pencil and enter the number in pen. Then I change it to pencil. At the end of the game this is recorded as an error. Is it supposed to be this way?
ReplyDeleteYes. Any entry you enter in pen that is incorrect will be recorded as an error (it cannot know if you meant to use pencil). Pencil marks will never record any errors.
ReplyDeleteExactly my problem too. Also, I'm a lousy typist and often enter "7" when I meant to enter "8". Even if I delete the "7" the instant I see that I mistyped it, it counts as an error.
ReplyDeleteWouldn't it be more fair if an error didn't count against you until you make your next entry? That way if you correct your own typo, or fail to change between pen and pencil but correct it immediately yourself, your score isn't based on bad keyboarding instead of bad sudokuing. (I know that's not a word, but I wish it was).
This comment section is forcing me to be anonymous. Not so. My email is: summers.mark@gmail.com
My high score has been 5874 ever since the "Well Done" box started displaying it. On 2/3/10, Derek said "The High Score is the highest score that you personally have ever scored on a puzzle." Today my high score dropped to 5189. How is that possible?
ReplyDeleteSteven
Steven, sorry I neglected to say that only the last 100 scores are kept on file. So your score of 5874 must have been more than 100 games ago and therefore has dropped off.
ReplyDeleteOn my last game, my score was above my average, yet my average went down quite a lot. What went wrong?
ReplyDeleteOnly the last 100 scores are kept on file. So what must have happened is that one or more of your older scores (clearly higher than your average) dropped off. Pruning is done once a day.
ReplyDeleteI got an 8699 score on one game. Is there a hall of fame? This game is great fun. I finaly found something to do during all this dang snow but if we get another 3 feet of snow I'll lose that great score. Please make a high score board or stop the snow I would love either. Thank you for a great game!
ReplyDeleteI just discovered your site, I love it! In the average rank index by age, does that mean the average score per age group? Or what is the index based on? Also, did you just create this site for the fun of it or is it a business for you?
ReplyDeleteThe average rank index by age is the average score for all players in that age group.
ReplyDeleteI created this site primarily for fun. The amount of money it earns is rather small. I could be doing other things and earning ten times the money but I want to do what's fun and give back to the community.
I have now changed the error recording to only happen if you make an entry in another cell before correcting the error. This solves the problem of entering an incorrect number by mistake.
ReplyDeleteIn my opinion, the emphasis on the time bonus has far too much weighting compared to accuracy and the penalty for using help tips.
ReplyDeleteI get the feeling that there are a (very small) number of players who are not playing fair. It's disheartening to see the same name at the top of the scoreboard with totally unbelievable scores. My suggestion is to make the window bigger that displays the final score after a game (covering the solution) and in that same window display where you rank at the moment. This would prevent the solution being displayed.
ReplyDeleteSudoku is a great. a great way to develop concentration. awesome scoring system
ReplyDeleteConcerns about a person "cheating" should be easy to address. If someone is keying in a solution that was determined outside the game, they will not key in the same order that someone uses to "solve" the puzzle. At most puzzles, there is a critical solution point. Also, there are numbers that only can be determined later, after many other boxes are filled in. Something more for you to think about, Derek.
ReplyDeleteThis post covers two unrelated issues:
ReplyDelete1) cheat prevention
2) mouse click detection
1) Quite a time ago (MAY 19, 2010 http://dkmgames.blogspot.com/2010/01/sudoku-scoring-system.html?showComment=1274277432457#c5691710041160704450 ) Jalair said: "If someone is keying in a solution that was determined outside the game, they will not key in the same order that someone uses to "solve" the puzzle."
Although Derek has explained why you can't use this for detecting cheating (JANUARY 17, 2010 http://dkmgames.blogspot.com/2010/01/sudoku-scoring-system.html?showComment=1263729667977#c2757534359714770073 ) I would like to comment on this.
Anyone with a well trained memory can solve any sudoku puzzle by heart, and I sometimes do so. When typing the solution from memory, I often work backward for a while, first entering the numbers that I found the last. Thus I would be catched by any cheat detector. Should solving a puzzle by heart be considered as cheating? I don't think so. On the other hand a cheat detector would not prevent real cheats. One could build a completely automated cheater using a screen grabber, an OCR program, a sudoku solver and some macros. Such an application could type the solution in a logical order.
And there is always the possibility that someone uses a solution scheme not foreseen by the programmer.
2) Quite often the focus in the puzzle seems frozen in the cell that was last filled in, ignoring mouse clicks in another cell, even after repeated clicking. Moving the focus with the arrow keys of the keyboard works always fine however. Replacing the USB mouse with another USB mouse did not make any difference. Do other users report the same problem?